Diversification of cropping systems is a lever for the management of epidemics. However, most research to date has focused on cultivar mixtures, especially for cereals, even though crop mixtures can also improve disease management. To investigate the benefits of crop mixtures, we studied the effect of different crop mixture characteristics (i.e., companion proportion, sowing date, and traits) on the protective effect of the mixture. We developed a SEIR (Susceptible, Exposed, Infectious, Removed) model of two damaging wheat diseases ( Zymoseptoria tritici and Puccinia triticina), which were applied to different canopy components, ascribable to wheat and a theoretical companion crop. We used the model to study the sensitivity of disease intensity to the following parameters: wheat-versus-companion proportion, companion sowing date and growth, and architectural traits. For both pathogens, the companion proportion had the strongest effect, with 25% of companion reducing disease severity by 50%. However, changing companion growth and architectural traits also significantly improved the protective effect. The effect of companion characteristics was consistent across different weather conditions. After decomposing the dilution and barrier effects, the model suggested that the barrier effect is maximized for an intermediate proportion of companion crop. Our study thus supports crop mixtures as a promising strategy to improve disease management. Future studies should identify real species and determine the combination of host and companion traits to maximize the protective effect of the mixture.